skills/cacaorick/skills/temp-folder/Gen Agent Trust Hub

temp-folder

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes filesystem commands including mkdir -p and rm -rf using variables ($TEMP_DIR) derived from local configuration files. Recursive deletion on variable-defined paths presents a risk of unintended data loss if the paths are manipulated.
  • Evidence: The cleanup section specifies the command rm -rf $TEMP_DIR/<用途名稱>.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by consuming instructions from project-level files to determine its operational logic and filesystem targets.
  • Ingestion points: The skill reads configuration from SPEC.md, AGENTS.md, CLAUDE.md, .agents.md, and rules files.
  • Boundary markers: None identified. The skill directly adopts paths found in these files.
  • Capability inventory: Filesystem creation and recursive deletion.
  • Sanitization: No sanitization or path validation logic is described to ensure the target directory resides within safe boundaries or the project root.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 12:25 AM